Development and validation of a control supervisor for UAV autonomous landing based in stereographic vision.
Authorship
P.B.G.
Master's Degree in Unmanned Aerial Systems
P.B.G.
Master's Degree in Unmanned Aerial Systems
Defense date
07.09.2026 13:15
07.09.2026 13:15
Summary
This Master’s thesis addresses the design and validation of a control supervisor for the autonomous landing of unmanned aerial vehicles (UAVs) based on stereo vision. The growing demand for robust and reliable autonomous systems makes the automation of critical maneuvers such as landing particularly relevant, where precision and safety are essential. The proposed system is integrated into a distributed architecture composed of a flight controller, an onboard computer, and a stereo camera, enabling environmental perception and real-time decision-making. Based on visual information, a perception system is developed to detect the landing area and estimate the UAV’s relative position with respect to it, leveraging stereo vision capabilities to obtain depth information. On this basis, a high-level control supervisor is designed as a finite state machine that manages the different phases of the landing process: search, detection, alignment, descent, touchdown, and abort. The transition logic between states allows adaptation to dynamic conditions and potential failures, improving system robustness. Validation is carried out in a simulation environment using tools such as Gazebo, PX4, and ROS, where different test scenarios and evaluation metrics are defined, including landing error, execution time, and system stability. Additionally, the system is integrated into real hardware, highlighting practical limitations related to latency, sensor noise, and computational constraints. The obtained results demonstrate the feasibility of the proposed approach, achieving precise and consistent landings under different conditions. Finally, system limitations are discussed, and future work directions are proposed, focusing on improving robustness and leveraging more powerful computing platforms.
This Master’s thesis addresses the design and validation of a control supervisor for the autonomous landing of unmanned aerial vehicles (UAVs) based on stereo vision. The growing demand for robust and reliable autonomous systems makes the automation of critical maneuvers such as landing particularly relevant, where precision and safety are essential. The proposed system is integrated into a distributed architecture composed of a flight controller, an onboard computer, and a stereo camera, enabling environmental perception and real-time decision-making. Based on visual information, a perception system is developed to detect the landing area and estimate the UAV’s relative position with respect to it, leveraging stereo vision capabilities to obtain depth information. On this basis, a high-level control supervisor is designed as a finite state machine that manages the different phases of the landing process: search, detection, alignment, descent, touchdown, and abort. The transition logic between states allows adaptation to dynamic conditions and potential failures, improving system robustness. Validation is carried out in a simulation environment using tools such as Gazebo, PX4, and ROS, where different test scenarios and evaluation metrics are defined, including landing error, execution time, and system stability. Additionally, the system is integrated into real hardware, highlighting practical limitations related to latency, sensor noise, and computational constraints. The obtained results demonstrate the feasibility of the proposed approach, achieving precise and consistent landings under different conditions. Finally, system limitations are discussed, and future work directions are proposed, focusing on improving robustness and leveraging more powerful computing platforms.
Direction
Iniesto Alba, María José (Tutorships)
Iniesto Alba, María José (Tutorships)
Court
Iniesto Alba, María José (Coordinator)
Iniesto Alba, María José (Chairman)
Arza García, Marcos (Secretary)
Fernandez Cabanas, Manés (Member)
Iniesto Alba, María José (Coordinator)
Iniesto Alba, María José (Chairman)
Arza García, Marcos (Secretary)
Fernandez Cabanas, Manés (Member)
Object Segmentation Using Aerial Image Analysis
Authorship
D.C.G.
Master's Degree in Unmanned Aerial Systems
D.C.G.
Master's Degree in Unmanned Aerial Systems
Defense date
07.09.2026 12:30
07.09.2026 12:30
Summary
Estimating vegetation coverage over large land areas is a task of great relevance in fields such as forest management, precision agriculture, and environmental monitoring. Traditional methods based on spectral indices or three dimensional reconstruction using LiDAR or Structure from Motion present limitations in terms of cost, operational complexity, or real time processing capability. This work proposes a complete system aimed at the automatic estimation of tree vegetation coverage area from aerial images captured by UAV. The adopted approach combines low cost photogrammetry with instance segmentation techniques based on deep learning, avoiding the need for three dimensional reconstructions or high cost active sensors. The development is structured around three main components: Construction and analysis of a photogrammetric dataset comprising 673 high resolution images with 1,611 annotated instances of dense tree vegetation, acquired by UAV over the Galicia region. Training of an instance segmentation model based on the YOLO26 architecture, achieving a precision of 0.84 and a recall of 0.82 on the test set, with a segmentation mAP50 of 0.89. Development and integration of a REST API using FastAPI, exposing the model in two operating modes: real time processing over RTSP video streams with result publication via MQTT, and offline analysis of individual images or ZIP compressed image sets. The results obtained demonstrate the technical feasibility of the system as an operational solution for vegetation coverage estimation, with a mean latency of 270 ms per frame on CPU and an area estimation error in the order of 10 15%, directly conditioned by the segmentation quality. The system provides a solid foundation for its evolution towards operational deployment on UAV platforms with onboard computing capabilities.
Estimating vegetation coverage over large land areas is a task of great relevance in fields such as forest management, precision agriculture, and environmental monitoring. Traditional methods based on spectral indices or three dimensional reconstruction using LiDAR or Structure from Motion present limitations in terms of cost, operational complexity, or real time processing capability. This work proposes a complete system aimed at the automatic estimation of tree vegetation coverage area from aerial images captured by UAV. The adopted approach combines low cost photogrammetry with instance segmentation techniques based on deep learning, avoiding the need for three dimensional reconstructions or high cost active sensors. The development is structured around three main components: Construction and analysis of a photogrammetric dataset comprising 673 high resolution images with 1,611 annotated instances of dense tree vegetation, acquired by UAV over the Galicia region. Training of an instance segmentation model based on the YOLO26 architecture, achieving a precision of 0.84 and a recall of 0.82 on the test set, with a segmentation mAP50 of 0.89. Development and integration of a REST API using FastAPI, exposing the model in two operating modes: real time processing over RTSP video streams with result publication via MQTT, and offline analysis of individual images or ZIP compressed image sets. The results obtained demonstrate the technical feasibility of the system as an operational solution for vegetation coverage estimation, with a mean latency of 270 ms per frame on CPU and an area estimation error in the order of 10 15%, directly conditioned by the segmentation quality. The system provides a solid foundation for its evolution towards operational deployment on UAV platforms with onboard computing capabilities.
Direction
Iniesto Alba, María José (Tutorships)
Iniesto Alba, María José (Tutorships)
Court
Iniesto Alba, María José (Coordinator)
Iniesto Alba, María José (Chairman)
Arza García, Marcos (Secretary)
Fernandez Cabanas, Manés (Member)
Iniesto Alba, María José (Coordinator)
Iniesto Alba, María José (Chairman)
Arza García, Marcos (Secretary)
Fernandez Cabanas, Manés (Member)
Economic impact of the entry of high speed in Galicia
Authorship
S.R.R.
Master's Degree in Planning and Territorial Management
S.R.R.
Master's Degree in Planning and Territorial Management
Defense date
07.14.2026 18:15
07.14.2026 18:15
Summary
The arrival of high speed rail in Galicia has led to a significant improvement in territorial connectivity, especially for the province of Ourense, the main entry point of this infrastructure into the region. This study analyzes the economic impact of high speed rail through the evolution of various indicators of mobility, economic activity, and tourism. The methodology used is based on a quantitative analysis of the temporal evolution of various indicators between 2015 and 2024. The main variable is the number of travelers on the Madrid Ourense line and the Ourense station. In addition, economic variables are analyzed, such as the Gross Added Value (GVA) of the services sector, the number of Social Security affiliations and the number of companies related to accommodation and hospitality, as well as tourist indicators linked to overnight stays and the accommodation supply. The results show a significant increase in rail mobility following the entry into operation of high speed services in 2021, accompanied by a positive trend in several economic and tourism indicators. However, although signs of economic dynamization can be observed, it is not possible to attribute all changes exclusively to the arrival of the new infrastructure, due to the influence of other external factors. In conclusion, high-speed rail has contributed to improving accessibility and strengthening the economic and tourism potential of Ourense, although its effects should be interpreted with caution.
The arrival of high speed rail in Galicia has led to a significant improvement in territorial connectivity, especially for the province of Ourense, the main entry point of this infrastructure into the region. This study analyzes the economic impact of high speed rail through the evolution of various indicators of mobility, economic activity, and tourism. The methodology used is based on a quantitative analysis of the temporal evolution of various indicators between 2015 and 2024. The main variable is the number of travelers on the Madrid Ourense line and the Ourense station. In addition, economic variables are analyzed, such as the Gross Added Value (GVA) of the services sector, the number of Social Security affiliations and the number of companies related to accommodation and hospitality, as well as tourist indicators linked to overnight stays and the accommodation supply. The results show a significant increase in rail mobility following the entry into operation of high speed services in 2021, accompanied by a positive trend in several economic and tourism indicators. However, although signs of economic dynamization can be observed, it is not possible to attribute all changes exclusively to the arrival of the new infrastructure, due to the influence of other external factors. In conclusion, high-speed rail has contributed to improving accessibility and strengthening the economic and tourism potential of Ourense, although its effects should be interpreted with caution.
Direction
GARCÍA ARIAS, ANA ISABEL (Tutorships)
GARCÍA ARIAS, ANA ISABEL (Tutorships)
Court
SANTE RIVEIRA, INES (Coordinator)
SANTE RIVEIRA, INES (Chairman)
VILA VAZQUEZ, JOSE IGNACIO (Secretary)
GARCÍA ARIAS, ANA ISABEL (Member)
SANTE RIVEIRA, INES (Coordinator)
SANTE RIVEIRA, INES (Chairman)
VILA VAZQUEZ, JOSE IGNACIO (Secretary)
GARCÍA ARIAS, ANA ISABEL (Member)
Social impact of urban rehabilitation projects in vulnerable neighborhoods: an analysis of the Sagrada Familia neighborhood A Coruna
Authorship
E.E.S.C.
Master's Degree in Planning and Territorial Management
E.E.S.C.
Master's Degree in Planning and Territorial Management
Defense date
07.14.2026 18:30
07.14.2026 18:30
Summary
This study analyzes the social impact of urban rehabilitation projects in the vulnerable neighborhood of Sagrada Familia A Coruna. Although these interventions have contributed to the improvement of infrastructure and public spaces through programs such as the ARI, they have not sufficiently incorporated the social dimensions necessary to ensure sustainable urban inclusion. A mixed-method approach is used combining documentary analysis, interviews, surveys, and statistical data to assess whether these interventions have fostered social cohesion or, on the contrary, generated processes of social fragmentation. The aim is to provide critical evidence to support the design of more inclusive and socially just urban rehabilitation policies in vulnerable urban contexts.
This study analyzes the social impact of urban rehabilitation projects in the vulnerable neighborhood of Sagrada Familia A Coruna. Although these interventions have contributed to the improvement of infrastructure and public spaces through programs such as the ARI, they have not sufficiently incorporated the social dimensions necessary to ensure sustainable urban inclusion. A mixed-method approach is used combining documentary analysis, interviews, surveys, and statistical data to assess whether these interventions have fostered social cohesion or, on the contrary, generated processes of social fragmentation. The aim is to provide critical evidence to support the design of more inclusive and socially just urban rehabilitation policies in vulnerable urban contexts.
Direction
VILA VAZQUEZ, JOSE IGNACIO (Tutorships)
VILA VAZQUEZ, JOSE IGNACIO (Tutorships)
Court
SANTE RIVEIRA, INES (Coordinator)
SANTE RIVEIRA, INES (Chairman)
GARCÍA ARIAS, ANA ISABEL (Secretary)
VILA VAZQUEZ, JOSE IGNACIO (Member)
SANTE RIVEIRA, INES (Coordinator)
SANTE RIVEIRA, INES (Chairman)
GARCÍA ARIAS, ANA ISABEL (Secretary)
VILA VAZQUEZ, JOSE IGNACIO (Member)